A Strategy of Extraction: How Deal-Making Is Displacing America’s Technology Competition Policy

In August 2025, the Trump administration struck a deal with Nvidia and AMD: the two chipmakers could resume selling advanced artificial intelligence (AI) semiconductors to the People’s Republic of China, provided that 15 percent of the revenue went to the U.S. Treasury. The president had initially demanded 20 percent, but settled for less after negotiations with Nvidia CEO Jensen Huang. By December, the administration upped the ante, greenlighting sales of Nvidia’s even more capable H200 chip to China at a 25 percent cut. In January 2026, the Bureau of Industry and Security formalized the shift, moving from a presumption of denial for advanced chip exports to China to a case-by-case review process.

The administration’s shifting approach to regulating advanced semiconductor sales to China reflects the inconsistencies that characterize U.S. strategic objectives related to technology competition. The same administration whose AI Action Plan claims to prioritize technological leadership is selling access to cutting-edge AI hardware, so long as Washington gets a cut of the proceeds. There is a legitimate debate about whether denying China access to advanced chips ultimately impedes Beijing’s progress or accelerates its drive toward self-sufficiency by guaranteeing a captive domestic market for indigenous chipmakers. But the administration is not in engaging that debate—it is auctioning access to the highest bidder. The United States does not currently have a coherent technology competition strategy, and the administration’s erratic, deal-driven approach is making it harder for firms to optimize research and development (R&D)  investment and strengthen supply chain resilience in ways that any serious competition with China requires.

Tools vs. Strategy

To understand the source of the United States’ strategic incoherence, it helps to look at the arc of technology policy across first Trump, Biden, and second Trump administrations. As I have argued elsewhere, analyzing economic tools alone obscures the strategic objectives they serve. All three administrations have used export controls, entity listings, and investment screening. But they have deployed these tools in service of fundamentally different visions.

The first Trump administration sought to weaponize interdependence within the existing rules-based order. Its landmark reforms to the Committee on Foreign Investment in the United States (CFIUS) and export control law (FIRRMA and ECRA, both 2018) were comprehensive, generally applied, and largely aligned with prevailing governance norms. Even aggressive actions like the Huawei entity listing were framed as security exceptions within a system of open exchange.

The Biden administration pursued technological supremacy through what might be called securitized embedded liberalism. It had a clearly articulated policy objective: maintain as large a lead as possible over China in critical technologies. National Security Advisor Jake Sullivan was the public face of this strategy, and the administration’s technology policies went through a full interagency vetting process. The resulting rules were rooted in established legal authorities and defined with painstaking specificity. The October 2022 semiconductor controls alone ran to hundreds of pages of detailed technical thresholds and compliance requirements. The administration updated them in October 2023and again in December 2024, issued an AI Diffusion Rule that created a three-tier framework for global chip exports, and finalized a connected vehicles rule banning Chinese and Russian software and hardware from the US auto supply chain. In total, the Bureau of Industry and Security (BIS) promulgated over 100 more rules under Biden than it had under the first Trump administration. Cumbersome as this approach was, it reflected a commitment to operating through regular order: clearly defined rules, transparent processes, and legal authorities that allies and firms alike could plan around.

The second Trump administration has broken both with Biden and its former incarnation. Its America First Investment Policy memorandum may sound, in title, like a continuation of the first term. But the memo is notably unconcerned with technology competition as a strategic objective. Where Biden’s team painstakingly defined which technologies mattered and why, the America First Investment Policy treats foreign investment primarily as tribute: what can other countries’ capital do for U.S. national power? It views foreign investors as political instruments of their home governments and shows little interest in governance principles like multilateral dialogue, reciprocity, or non-discrimination. The administration’s July 2025 AI Action Plan nominally addresses technology competition, but its substance reveals the same logic. Pillar I emphasizes deregulation to benefit the major technology firms that the administration courted during the presidential campaign. Pillar III, on international diplomacy and security, pledges to strengthen export controls and allied coordination. In practice, however, it focuses on pressuring allies to adopt American AI systems rather than developing genuine cooperation on innovation or enforcement, and the administration’s own dealmaking directly contradicts any commitment to strict controls. I have characterized this posture as expansionary mercantilism: a return to viewing economic exchange as a tool of state domination rather than a source of mutual prosperity. But the evidence of the past year suggests that even mercantilism does not fully capture what is happening.

Beyond Mercantilism

What, then, is driving the administration’s technology policy? Alexander Cooley and Daniel Nexon offer a useful framework. In a recent Foreign Affairs essay, they argue that the administration’s foreign policy is best understood as a system in which private enrichment of regime insiders is not a side effect of dealmaking but its central purpose. Many news reports treat the private financial arrangements surrounding Trump’s foreign policy as side payments struck in service of a larger agreement. Cooley and Nexon contend the opposite: the deals are themselves the strategy. The administration practices what they call “transactional bundling,” rolling together diplomatic concessions, investment pledges, and private financial arrangements that benefit the president’s inner circle into opaque megapackages.

Transactional bundling extends to technology policy. For example, in May 2025, Trump envoy Steve Witkoff brokered an agreement to lift U.S. restrictions on the export of advanced AI chips to a United Arab Emirates (UAE) technology firm chaired by Sheikh Tahnoon bin Zayed Al Nahyan, the country’s national security advisor. The Biden administration had restricted this access on national security grounds, requiring the company to divest from Huawei hardware and accept stringent security conditions. Around the same time, MGX, an investment firm also chaired by Tahnoon, purchased $2 billion worth of stablecoin from World Liberty Financial, a cryptocurrency venture whose co-founders include Witkoff, his two sons, Trump, and Trump’s three sons.

The same logic that produces these arrangements also requires dismantling the institutions that would constrain them. The administration has gutted the bureaucratic infrastructure that previous administrations built to wage the technology competition. In January 2026, the administration pushed out the executive director of the Office of Information and Communications Technology and Services office at BIS, the unit responsible for the connected vehicles rule and other critical supply chain security initiatives. The Commerce Department also dropped planned restrictions on Chinese drones. Key staff across BIS subsequently departed or have been reassigned.

This de-institutionalization is not accidental. A functioning technocratic bureaucracy pursuing a coherent technology supremacy strategy is an obstacle to the kind of dealmaking the administration favors. National security commitments, once established through formal regulatory processes, cannot easily be traded away in bilateral negotiations. They constrain the administration’s bargaining space. By hollowing out the agencies that make and enforce these commitments, the administration preserves its freedom to negotiate ad hoc arrangements. The connected vehicles rule is instructive: it remains formally on the books, but the institutional capacity to enforce and extend it has been gutted.

Policy Unpredictability And Firm Behavior

The consequences for U.S. technology firms are significant. When the regulatory framework is stable, firms invest in compliance, R&D, and long-term strategy. When it is unpredictable and subject to negotiation, the strategic calculus shifts. The marginal dollar moves from research to lobbying because a policy environment shaped by presidential discretion is one where political access determines market access. Nvidia’s federal lobbying expenditures increased nearly sevenfold between 2024 and 2025, while total U.S. lobbying spending hit a record $5.08 billion in the first year of Trump’s second term.

The January 2026 shift to case-by-case review at BIS intensified this dynamic. Every export transaction is now a negotiation rather than an application of clear rules. Firms that cultivate relationships with the administration can expect favorable treatment. Those that do not face uncertainty. The Nvidia deal itself set the precedent: as one analyst noted, the revenue-sharing arrangement is a “very slippery slope” that other industries will soon be asked to slide down.

This is corrosive to the innovation ecosystem the United States needs to maintain its technology lead. As Miscik, Orszag, and Bunzel have warned, discretionary state capitalism risks concentrating gains among politically connected incumbents while sapping innovation. The whole point of the Biden-era CHIPS and Science Act was to channel investment into domestic semiconductor manufacturing and R&D. A policy environment that rewards political proximity over productive investment works at cross-purposes with that goal.

Plurilateral Breakdown

For allied and partner countries, the picture is equally troubling. The rescission of Biden’s AI Diffusion Rule on May 13, 2025, two days before its compliance deadline, eliminated the predictable, tiered framework that allies had been preparing to operate within. In its place, the administration has pursued bilateral negotiations in which each country must individually negotiate its access to advanced AI chips.

The Gulf states were first: the UAE secured access to 500,000 advanced Nvidia chips annually; Saudi Arabia received a comparable arrangement bundled into a broader $600 billion investment package. But this model is now being extended to close allies. South Korea, which had enjoyed preferential Tier 1 status under Biden’s framework, now faces conditions for chip access that include investing in U.S. data centers, with Commerce Department officials citing the Gulf deals as the template.

This approach transforms chip exports from a domain governed by allies’ shared strategic objectives into a source of bilateral leverage. And the administration has shown a willingness to use economic tools coercively in entirely unrelated contexts. On March 3, 2026, Trump threatened to cut off all trade with Spain after it refused to allow the U.S. military use of joint bases for strikes on Iran. If tariff policy can be weaponized over basing rights, allied governments have every reason to fear that technology dependencies will be similarly exploited. As Stephen Walt has argued, a predatory hegemon that exploits partners’ dependencies drives them to seek alternatives.

The implications for multilateral technology cooperation are severe. As Emily Kilcrease and I argued when the October 2022 controls were first announced, unilateral export controls cannot succeed without allied coordination. The semiconductor supply chain’s network topography, even with its chokepoints, is far less hierarchical than the U.S. dollar system and therefore much harder to unilaterally control. The Netherlands expanded its semiconductor export controls in April 2025, restricting sales of advanced ASML lithography equipment. But nearly 30 percent of ASML’s sales go to Chinese customers. Why would the Dutch, or the Japanese, continue to sacrifice market access to enforce a technology denial strategy that Washington itself is undermining through revenue-sharing deals and ad hoc bilateral exemptions?

The Consequences Of Extraction

In the past, the United States and its allies worked toward a plurilateral technology ecosystem rooted in market democracies that could offer an alternative to China’s state-directed tech stack. That project required trust, institutional infrastructure, and shared rules. All three are now eroding.

China is well placed to capitalize on Washington’s current disinterest in forward-looking national security strategy. Beijing has spent decades building its indigenous semiconductor capabilities precisely because it anticipated that the United States would use its dominant technological positions coercively. The current administration’s behavior validates that assumption while simultaneously weakening the coalition that made its technological chokepoints effective sources of national security. Capable people within the administration may genuinely want a coherent technology competition strategy. But they are operating within a system whose primary logic is extractive, not strategic.

A tool is not a strategy, and extraction is not strategic competition. The United States still holds formidable positions in the global semiconductor supply chain. But those positions depend on sustained investment, allied cooperation, and institutional credibility. The question confronting Indo-Pacific policymakers is not simply what Washington will do on chips and AI, but whether the United States remains a partner capable of credibly committing to a shared approach to technology governance. The parallel question for U.S. policymakers and strategists is what it will take to rebuild the trust required for allied cooperation over technology policy in future years, and how much advantage will be lost in the meantime.

About the author